Flight Position Predictor

As part of the “Flight Operations Engineering” traject of my studies in Aviation, an assignment was given to develop a model that could predict the future position of an aircraft on a flight. Achieving this involved processing many dataset, and developing multiple models for the prediction.

To process the data, use was made of many Python libraries, namely, Avipy, pandas, numpy, scikit-learn and tensorflow (this one was my favorite to tinker with). The datasets came in the form of csv files containing ADS-B data of many historic flights on the Barcelona - Amsterdam route. To make use of this data for the training of a model, techniques such as data cleaning, outlier analysis and data scaling was used.

Multiple models were used and evaluated to predict for the prediction:

• Linear Regression
• Random Forest Regression
• Artificial Neural Network (ANN)
• Recurrent Neural Network (RNN)

Among these, the RNN model outperformed the others by a great margin, due to its ability to make accurate predictions in a time series dataset.